Comparison of several satellite-derived Sun-Induced Fluorescence products

Large uncertainties remain in our representation of the global carbon budget, in particular regarding the spatial and temporal dynamics of the net land surface CO2 fluxes along with its two constitutive components, photosynthesis and respiration. Bolstered by the evidenced linear relationship between remotely sensed sun-induced fluorescence (SIF) and plant gross carbon uptake (GPP - gross primary productivity) at broad spatial and temporal scales, satellite SIF products are foreseen to provide significant constraint on one of the key component of the terrestrial carbon cycle, and ultimately to help reducing the uncertainties in the projections of the fate of carbon sinks and sources under a changing climate.Global SIF estimates are now "routinely" produced from observations of space-borne spectrometers having sufficient spectral resolution/sampling in solar Fraunhofer lines or atmospheric absorption bands in the visible - near-infrared domain. Differences between SIF products derived from different instruments are expected depending on evaluated wavelengths (SIF has a spectral signature with maxima around 685 and 740 nm) and their own observation characteristics (time of satellite overpass, spatial resolution, revisit frequency, spectral resolution, etc.). For instance, SIF products estimated at 760 nm (GOSAT, OCO-2) are about 1.5 times lower than estimates at 740 nm (GOME-2, SCIAMACHY). However, as highlighted by Köhler et al. (2015), strong discrepancies in SIF absolute values may arise for products derived from the same set of observations (GOME-2) but using different estimation algorithms. In the view of using satellite SIF products to calibrate terrestrial biosphere models (e.g. through data assimilation), this is highly problematic, especially for evergreen ecosystems where SIF magnitude is the only observational constraint that can be made use of.In this study, we compare several gridded satellite SIF products and quantify their similarities/discrepancies with respect to both their absolute value and seasonality (plant phenology): GOME-2, OCO2, GOSAT, and SCIAMACHY. Our main objective is to assess the potential impacts of their differences in a data assimilation perspective.